Aphasia, an impairment in language processing, is a common disorder associated with stroke. The hallmark deficit of aphasia is the inability to name objects or people (anomia). Many patients with chronic aphasia can improve with naming therapy, but treatment is not universally effective. If the reason for differences in treatment effectiveness were better understood, an instrument could be developed to identify the patients who have the greatest potential to improve. Furthermore, a better understanding of the predictors for language recovery could provide crucial information about the mechanisms supporting brain plasticity during rehabilitation. Our group has previously demonstrated that naming improvements due to therapy are primarily associated with functional modulation of the cerebral cortex in the left hemisphere. It follows that structural damage to cortical regions in th left hemisphere is a limiting factor for recovery. Nonetheless, preservation of the cerebral cortex cannot fully predict therapy-induced recovery. Some patients with apparently intact cortical structures are not able to recruit these areas during therapy and fail to improve. This apparent inconsistency may be related to limitations in brain mapping techniques and our hitherto inability to define the extent and location of brain damage after stroke. Specifically, cortical regions may be disconnected as a result of white matter loss. Conventional assessment tools underappreciate cortical disconnection, but it likely plays an important role in naming recovery because it prevents recruitment of spared cortical areas during therapy. With new methodological improvements in brain mapping, this hypothesis can be directly tested. It is now possible to chart neural connections in the entire brain (the brain connectome) using magnetic resonance diffusion tensor imaging. For this project, we developed optimized connectome-mapping techniques to assess neural connectivity in patients with aphasia due to a previous stroke. We aim to investigate the impacts of cortical necrosis and cortical disconnection on chronic naming impairments and treatment-induced naming recovery. This proposal leverages high-quality imaging and behavioral data from a large prospective treatment trial in aphasia to accomplish our goals in a cost-effective manner. We will assess specific regions in the left hemisphere that have been associated with lexical-semantic retrieval and phonological processing during naming. We will also create a clinical scale for the prediction of treatment outcome based on a personalized assessment of cortical damage and cortical connectivity, advocating for each patient to be treated according to their individualized brain network profile.